Today I learned: Thrust zip_iterator is not Boost zip_iterator

Unfortunately, if you’re used to sorting zipped up ranges with Thrust, that does not translate into Boost.

I - I had so much love for Boost. But then I learned, Nvidia… They did it better. I’m not sure what to believe anymore.

(this is kind of a joke thread. But seriously, Boost’s iterator adapters are not as nice, it seems, as Thrust’s)

Someday C++17 or C++21 will make everything perfect.

I am so not looking forward to 1000+ pages of language specification.

[duplicate removed]

i’m not tried but may be boost.range or Range3 can help?

I did think about Ranges, actually :P

But I think I was too expectant with my Thrust experience transferring into Boost literally.

I’m less sad now.

I’m not curious what C++17 would bring to the table for CUDA. I know Concepts didn’t make it in. Nor modules.

I think structured bindings did and that’d make working with tuples significantly easier. Some C++14 support would be nice in CUDA as well. The ability to just auto func_name() and have the return be deduced by the return expression would be a boon though it’s not necessary for writing good software.